Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

data_dir = './data'
import helper
show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x1bfbb6f5eb8>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x1bfbc36a940>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.0.1
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    real_input = tf.placeholder(dtype=tf.float32, shape=(None, image_height, image_width, image_channels), name='input_real')
    z_data = tf.placeholder(dtype=tf.float32, shape=(None, z_dim), name='input_z')
    learn_rate = tf.placeholder(dtype=tf.float32)
    return real_input, z_data, learn_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the generator, tensor logits of the generator).

def discriminator(images, reuse=False, alpha=0.2):
    """
    Create the discriminator network
    :param image: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # TODO: Implement Function
    with tf.variable_scope('discriminator', reuse=reuse):
        
        # Input layer is 28s28x(1 or 3)
        x1 = tf.layers.conv2d(images, filters=64, kernel_size=4, strides=2, padding='valid') 
        relu1 = tf.maximum(alpha * x1, x1)
        # 13x13x64
        
        x2 = tf.layers.conv2d(relu1, filters=128, kernel_size=3, strides=2, padding='valid') 
        bn2 = tf.layers.batch_normalization(x2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 6x6x128
        
        x3 = tf.layers.conv2d(relu2, filters=256, kernel_size=2, strides=2, padding='valid') 
        bn3 = tf.layers.batch_normalization(x3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        # 3x3x256

        # Flatten it
        flat = tf.reshape(relu3, (-1, 3*3*256))       
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)

    return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variabes in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True, alpha=0.2):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    
    with tf.variable_scope('generator', reuse=not is_train):
        
        # First fully connected layer
        x1 = tf.layers.dense(z, 3*3*512)
        
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 3, 3, 512))
        
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        # 3x3x512 now
        
        x2 = tf.layers.conv2d_transpose(x1, filters=256, kernel_size=2, strides=2, padding='valid') 
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        # 6x6x256 now
        
        x3 = tf.layers.conv2d_transpose(x2, filters=128, kernel_size=3, strides=2, padding='valid') 
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # 13x13x128 now
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, kernel_size=4, strides=2, padding='valid') 
        # 28x28x3 now
        
        out = tf.tanh(logits)
        
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim, alpha=0.2):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    # TODO: Implement Function
    g_model = generator(input_z, out_channel_dim, is_train=True)
    
    d_model_real, d_logits_real = discriminator(input_real, reuse=False)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)
    
    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real) * 0.9))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation)

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    # Get weights and bias to update
    t_vars = tf.trainable_variables()
    
    d_vars = [var for var in t_vars if var.name.startswith("discriminator")]
    g_vars = [var for var in t_vars if var.name.startswith("generator")]
    
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1 = beta1).minimize(d_loss, var_list = d_vars)
    
    g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1 = beta1).minimize(g_loss, var_list = g_vars)
    
    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    #tf.reset_default_graph()
    steps = 0
    print_every = 4
    show_every = 40
    lr_alpha = 0.2 #leaky relu slope
    
    
#     img_channels = 0
#     if data_image_mode is "RGB":
#         img_channels = 3
#     if data_image_mode is "L":
#         img_channels = 1
        
#     print("Data shape: {}".format(*data_shape))
    
    width = data_shape[1]
    height = data_shape[2]
    channels = data_shape[3]
    
    
    input_real, input_z, learn_rate = model_inputs(width, height, channels, z_dim)
    d_loss, g_loss = model_loss(input_real, input_z,channels, alpha=lr_alpha)
    d_opt, g_opt = model_opt(d_loss, g_loss, learn_rate, beta1)
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                image_input = 2 * batch_images #batch images is in range -0.5 to 0.5, so multiply by 2 to rescale
                steps += 1

                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                # Run optimizers
                _ = sess.run(d_opt, feed_dict={input_real: image_input, input_z: batch_z, learn_rate: learning_rate})
                '''
                Run the generator optimizer TWICE to avoid fast convergence of the discriminator network
                '''
                _ = sess.run(g_opt, feed_dict={input_real: image_input, input_z: batch_z, learn_rate: learning_rate})
                _ = sess.run(g_opt, feed_dict={input_real: image_input, input_z: batch_z, learn_rate: learning_rate})
                
                if steps % print_every == 0:
                    # At the end of each epoch, get the losses and print them out
                    train_loss_d = sess.run(d_loss, {input_z: batch_z, input_real: image_input, learn_rate: learning_rate })
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epochs),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    # Save losses to view after training
                    #losses.append((train_loss_d, train_loss_g))

                if steps % show_every == 0:
                    n_images = 9
                    show_generator_output(sess, n_images, input_z, channels, data_image_mode)
    print("-------Done-------")

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 128
z_dim = 200
learning_rate = 0.0002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 5

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))

with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/5... Discriminator Loss: 3.4581... Generator Loss: 0.0389
Epoch 1/5... Discriminator Loss: 3.4230... Generator Loss: 0.0429
Epoch 1/5... Discriminator Loss: 3.5303... Generator Loss: 0.0417
Epoch 1/5... Discriminator Loss: 3.3557... Generator Loss: 0.0479
Epoch 1/5... Discriminator Loss: 3.0844... Generator Loss: 0.0638
Epoch 1/5... Discriminator Loss: 2.9455... Generator Loss: 0.0776
Epoch 1/5... Discriminator Loss: 2.7566... Generator Loss: 0.1040
Epoch 1/5... Discriminator Loss: 2.4104... Generator Loss: 0.1444
Epoch 1/5... Discriminator Loss: 2.6959... Generator Loss: 0.1125
Epoch 1/5... Discriminator Loss: 2.2306... Generator Loss: 0.1988

Epoch 1/5... Discriminator Loss: 2.1911... Generator Loss: 0.2074
Epoch 1/5... Discriminator Loss: 2.4010... Generator Loss: 0.1957
Epoch 1/5... Discriminator Loss: 2.2326... Generator Loss: 0.2477
Epoch 1/5... Discriminator Loss: 2.1811... Generator Loss: 0.2628
Epoch 1/5... Discriminator Loss: 2.4010... Generator Loss: 0.2593
Epoch 1/5... Discriminator Loss: 2.3193... Generator Loss: 0.2690
Epoch 1/5... Discriminator Loss: 2.4496... Generator Loss: 0.2960
Epoch 1/5... Discriminator Loss: 2.0807... Generator Loss: 0.3651
Epoch 1/5... Discriminator Loss: 1.9878... Generator Loss: 0.5094
Epoch 1/5... Discriminator Loss: 1.6766... Generator Loss: 0.5962

Epoch 1/5... Discriminator Loss: 1.6042... Generator Loss: 0.5727
Epoch 1/5... Discriminator Loss: 1.4442... Generator Loss: 0.6982
Epoch 1/5... Discriminator Loss: 1.3599... Generator Loss: 0.6906
Epoch 1/5... Discriminator Loss: 1.9218... Generator Loss: 0.4469
Epoch 1/5... Discriminator Loss: 1.5258... Generator Loss: 0.5777
Epoch 1/5... Discriminator Loss: 1.4659... Generator Loss: 0.5617
Epoch 1/5... Discriminator Loss: 1.8414... Generator Loss: 0.5134
Epoch 1/5... Discriminator Loss: 1.5356... Generator Loss: 0.6260
Epoch 1/5... Discriminator Loss: 1.5200... Generator Loss: 0.6376
Epoch 1/5... Discriminator Loss: 1.7144... Generator Loss: 0.5250

Epoch 1/5... Discriminator Loss: 1.6163... Generator Loss: 0.5376
Epoch 1/5... Discriminator Loss: 1.4552... Generator Loss: 0.6348
Epoch 1/5... Discriminator Loss: 1.5285... Generator Loss: 0.6167
Epoch 1/5... Discriminator Loss: 1.5345... Generator Loss: 0.5781
Epoch 1/5... Discriminator Loss: 1.4738... Generator Loss: 0.6118
Epoch 1/5... Discriminator Loss: 1.4934... Generator Loss: 0.6006
Epoch 1/5... Discriminator Loss: 1.5007... Generator Loss: 0.6125
Epoch 1/5... Discriminator Loss: 1.4521... Generator Loss: 0.6420
Epoch 1/5... Discriminator Loss: 1.4768... Generator Loss: 0.6000
Epoch 1/5... Discriminator Loss: 1.5391... Generator Loss: 0.5720

Epoch 1/5... Discriminator Loss: 1.4963... Generator Loss: 0.5805
Epoch 1/5... Discriminator Loss: 1.5420... Generator Loss: 0.5865
Epoch 1/5... Discriminator Loss: 1.4705... Generator Loss: 0.6176
Epoch 1/5... Discriminator Loss: 1.4613... Generator Loss: 0.6127
Epoch 1/5... Discriminator Loss: 1.4276... Generator Loss: 0.5938
Epoch 1/5... Discriminator Loss: 1.4413... Generator Loss: 0.6006
Epoch 1/5... Discriminator Loss: 1.4327... Generator Loss: 0.6325
Epoch 1/5... Discriminator Loss: 1.4225... Generator Loss: 0.6158
Epoch 1/5... Discriminator Loss: 1.4504... Generator Loss: 0.6345
Epoch 1/5... Discriminator Loss: 1.3685... Generator Loss: 0.6676

Epoch 1/5... Discriminator Loss: 1.3705... Generator Loss: 0.6529
Epoch 1/5... Discriminator Loss: 1.3758... Generator Loss: 0.6286
Epoch 1/5... Discriminator Loss: 1.3675... Generator Loss: 0.6474
Epoch 1/5... Discriminator Loss: 1.4032... Generator Loss: 0.6442
Epoch 1/5... Discriminator Loss: 1.4066... Generator Loss: 0.6314
Epoch 1/5... Discriminator Loss: 1.3693... Generator Loss: 0.6488
Epoch 1/5... Discriminator Loss: 1.3427... Generator Loss: 0.6786
Epoch 1/5... Discriminator Loss: 1.3876... Generator Loss: 0.6276
Epoch 1/5... Discriminator Loss: 1.3828... Generator Loss: 0.6145
Epoch 1/5... Discriminator Loss: 1.3610... Generator Loss: 0.6131

Epoch 1/5... Discriminator Loss: 1.3336... Generator Loss: 0.6447
Epoch 1/5... Discriminator Loss: 1.2955... Generator Loss: 0.6904
Epoch 1/5... Discriminator Loss: 1.3198... Generator Loss: 0.6531
Epoch 1/5... Discriminator Loss: 1.2947... Generator Loss: 0.7093
Epoch 1/5... Discriminator Loss: 1.2717... Generator Loss: 0.6889
Epoch 1/5... Discriminator Loss: 1.2733... Generator Loss: 0.6493
Epoch 1/5... Discriminator Loss: 1.2760... Generator Loss: 0.6531
Epoch 1/5... Discriminator Loss: 1.3038... Generator Loss: 0.6807
Epoch 1/5... Discriminator Loss: 1.3066... Generator Loss: 0.6625
Epoch 1/5... Discriminator Loss: 1.3148... Generator Loss: 0.6130

Epoch 1/5... Discriminator Loss: 1.2506... Generator Loss: 0.6952
Epoch 1/5... Discriminator Loss: 1.2497... Generator Loss: 0.7037
Epoch 1/5... Discriminator Loss: 1.2579... Generator Loss: 0.6656
Epoch 1/5... Discriminator Loss: 1.1900... Generator Loss: 0.7292
Epoch 1/5... Discriminator Loss: 1.2126... Generator Loss: 0.7008
Epoch 1/5... Discriminator Loss: 1.2654... Generator Loss: 0.6950
Epoch 1/5... Discriminator Loss: 1.2347... Generator Loss: 0.6871
Epoch 1/5... Discriminator Loss: 1.2678... Generator Loss: 0.7452
Epoch 1/5... Discriminator Loss: 1.2330... Generator Loss: 0.6576
Epoch 1/5... Discriminator Loss: 1.2094... Generator Loss: 0.7097

Epoch 1/5... Discriminator Loss: 1.2210... Generator Loss: 0.7551
Epoch 1/5... Discriminator Loss: 1.2332... Generator Loss: 0.7314
Epoch 1/5... Discriminator Loss: 1.2023... Generator Loss: 0.7063
Epoch 1/5... Discriminator Loss: 1.2345... Generator Loss: 0.6666
Epoch 1/5... Discriminator Loss: 1.2092... Generator Loss: 0.8157
Epoch 1/5... Discriminator Loss: 1.1440... Generator Loss: 0.7961
Epoch 1/5... Discriminator Loss: 1.2171... Generator Loss: 0.6195
Epoch 1/5... Discriminator Loss: 1.1903... Generator Loss: 0.7071
Epoch 1/5... Discriminator Loss: 1.2156... Generator Loss: 0.7063
Epoch 1/5... Discriminator Loss: 1.1636... Generator Loss: 0.8903

Epoch 1/5... Discriminator Loss: 1.2434... Generator Loss: 0.5674
Epoch 1/5... Discriminator Loss: 1.1708... Generator Loss: 0.6788
Epoch 1/5... Discriminator Loss: 1.2465... Generator Loss: 0.7767
Epoch 1/5... Discriminator Loss: 1.1666... Generator Loss: 0.7224
Epoch 1/5... Discriminator Loss: 1.2945... Generator Loss: 0.5305
Epoch 1/5... Discriminator Loss: 1.2412... Generator Loss: 0.7259
Epoch 1/5... Discriminator Loss: 1.1828... Generator Loss: 0.7124
Epoch 1/5... Discriminator Loss: 1.2463... Generator Loss: 0.8414
Epoch 1/5... Discriminator Loss: 1.1908... Generator Loss: 0.8755
Epoch 1/5... Discriminator Loss: 1.2126... Generator Loss: 0.6991

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 128
z_dim = 200
learning_rate = 0.002
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 5

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/5... Discriminator Loss: 5.5456... Generator Loss: 0.0078
Epoch 1/5... Discriminator Loss: 5.0598... Generator Loss: 0.0146
Epoch 1/5... Discriminator Loss: 3.9753... Generator Loss: 0.0740
Epoch 1/5... Discriminator Loss: 3.8723... Generator Loss: 0.1905
Epoch 1/5... Discriminator Loss: 3.0583... Generator Loss: 0.4933
Epoch 1/5... Discriminator Loss: 2.5308... Generator Loss: 0.6966
Epoch 1/5... Discriminator Loss: 2.3367... Generator Loss: 0.6699
Epoch 1/5... Discriminator Loss: 2.2866... Generator Loss: 0.6561
Epoch 1/5... Discriminator Loss: 1.9035... Generator Loss: 0.6272
Epoch 1/5... Discriminator Loss: 2.2904... Generator Loss: 0.3871

Epoch 1/5... Discriminator Loss: 2.4016... Generator Loss: 0.4010
Epoch 1/5... Discriminator Loss: 2.2389... Generator Loss: 0.3237
Epoch 1/5... Discriminator Loss: 2.0744... Generator Loss: 0.4194
Epoch 1/5... Discriminator Loss: 2.2069... Generator Loss: 0.3274
Epoch 1/5... Discriminator Loss: 2.1764... Generator Loss: 0.4342
Epoch 1/5... Discriminator Loss: 1.7637... Generator Loss: 0.4268
Epoch 1/5... Discriminator Loss: 1.9803... Generator Loss: 0.4478
Epoch 1/5... Discriminator Loss: 1.8833... Generator Loss: 0.4877
Epoch 1/5... Discriminator Loss: 1.7826... Generator Loss: 0.5138
Epoch 1/5... Discriminator Loss: 2.1280... Generator Loss: 0.4153

Epoch 1/5... Discriminator Loss: 1.8166... Generator Loss: 0.4748
Epoch 1/5... Discriminator Loss: 1.7872... Generator Loss: 0.5130
Epoch 1/5... Discriminator Loss: 1.9392... Generator Loss: 0.4516
Epoch 1/5... Discriminator Loss: 1.9639... Generator Loss: 0.4566
Epoch 1/5... Discriminator Loss: 1.7967... Generator Loss: 0.5056
Epoch 1/5... Discriminator Loss: 1.8183... Generator Loss: 0.5074
Epoch 1/5... Discriminator Loss: 1.8650... Generator Loss: 0.5214
Epoch 1/5... Discriminator Loss: 1.8331... Generator Loss: 0.5289
Epoch 1/5... Discriminator Loss: 1.9835... Generator Loss: 0.4517
Epoch 1/5... Discriminator Loss: 1.7936... Generator Loss: 0.4472

Epoch 1/5... Discriminator Loss: 2.2470... Generator Loss: 0.3921
Epoch 1/5... Discriminator Loss: 1.8150... Generator Loss: 0.4327
Epoch 1/5... Discriminator Loss: 1.6977... Generator Loss: 0.5097
Epoch 1/5... Discriminator Loss: 1.9749... Generator Loss: 0.4090
Epoch 1/5... Discriminator Loss: 1.7615... Generator Loss: 0.5161
Epoch 1/5... Discriminator Loss: 1.8771... Generator Loss: 0.5027
Epoch 1/5... Discriminator Loss: 1.6848... Generator Loss: 0.5168
Epoch 1/5... Discriminator Loss: 1.6390... Generator Loss: 0.5100
Epoch 1/5... Discriminator Loss: 1.7868... Generator Loss: 0.5569
Epoch 1/5... Discriminator Loss: 1.6350... Generator Loss: 0.4983

Epoch 1/5... Discriminator Loss: 1.6769... Generator Loss: 0.6154
Epoch 1/5... Discriminator Loss: 1.9744... Generator Loss: 0.4220
Epoch 1/5... Discriminator Loss: 1.5967... Generator Loss: 0.5426
Epoch 1/5... Discriminator Loss: 1.7262... Generator Loss: 0.4654
Epoch 1/5... Discriminator Loss: 1.6146... Generator Loss: 0.5578
Epoch 1/5... Discriminator Loss: 1.6083... Generator Loss: 0.5584
Epoch 1/5... Discriminator Loss: 1.7457... Generator Loss: 0.5042
Epoch 1/5... Discriminator Loss: 1.6491... Generator Loss: 0.5666
Epoch 1/5... Discriminator Loss: 1.7519... Generator Loss: 0.5609
Epoch 1/5... Discriminator Loss: 1.8817... Generator Loss: 0.4466

Epoch 1/5... Discriminator Loss: 1.7964... Generator Loss: 0.4796
Epoch 1/5... Discriminator Loss: 1.8777... Generator Loss: 0.4124
Epoch 1/5... Discriminator Loss: 1.6268... Generator Loss: 0.5208
Epoch 1/5... Discriminator Loss: 1.7379... Generator Loss: 0.5720
Epoch 1/5... Discriminator Loss: 1.6986... Generator Loss: 0.4985
Epoch 1/5... Discriminator Loss: 1.7949... Generator Loss: 0.4625
Epoch 1/5... Discriminator Loss: 1.5232... Generator Loss: 0.5991
Epoch 1/5... Discriminator Loss: 1.7911... Generator Loss: 0.4519
Epoch 1/5... Discriminator Loss: 1.8248... Generator Loss: 0.4564
Epoch 1/5... Discriminator Loss: 1.6106... Generator Loss: 0.5230

Epoch 1/5... Discriminator Loss: 2.1224... Generator Loss: 0.4179
Epoch 1/5... Discriminator Loss: 1.9737... Generator Loss: 0.4601
Epoch 1/5... Discriminator Loss: 2.0225... Generator Loss: 0.4022
Epoch 1/5... Discriminator Loss: 1.7702... Generator Loss: 0.4698
Epoch 1/5... Discriminator Loss: 1.9366... Generator Loss: 0.6245
Epoch 1/5... Discriminator Loss: 2.0964... Generator Loss: 0.4019
Epoch 1/5... Discriminator Loss: 1.8506... Generator Loss: 0.3959
Epoch 1/5... Discriminator Loss: 1.6972... Generator Loss: 0.4786
Epoch 1/5... Discriminator Loss: 1.6880... Generator Loss: 0.5811
Epoch 1/5... Discriminator Loss: 1.7495... Generator Loss: 0.5281

Epoch 1/5... Discriminator Loss: 1.5747... Generator Loss: 0.6096
Epoch 1/5... Discriminator Loss: 1.6141... Generator Loss: 0.5237
Epoch 1/5... Discriminator Loss: 1.5931... Generator Loss: 0.5725
Epoch 1/5... Discriminator Loss: 1.4896... Generator Loss: 0.6675
Epoch 1/5... Discriminator Loss: 1.5834... Generator Loss: 0.6293
Epoch 1/5... Discriminator Loss: 1.7446... Generator Loss: 0.5716
Epoch 1/5... Discriminator Loss: 1.7202... Generator Loss: 0.4914
Epoch 1/5... Discriminator Loss: 1.5433... Generator Loss: 0.6159
Epoch 1/5... Discriminator Loss: 1.5036... Generator Loss: 0.7117
Epoch 1/5... Discriminator Loss: 1.7254... Generator Loss: 0.5457

Epoch 1/5... Discriminator Loss: 1.6537... Generator Loss: 0.5886
Epoch 1/5... Discriminator Loss: 1.6356... Generator Loss: 0.5459
Epoch 1/5... Discriminator Loss: 1.7143... Generator Loss: 0.4795
Epoch 1/5... Discriminator Loss: 1.5915... Generator Loss: 0.5910
Epoch 1/5... Discriminator Loss: 1.6438... Generator Loss: 0.6069
Epoch 1/5... Discriminator Loss: 1.5688... Generator Loss: 0.5417
Epoch 1/5... Discriminator Loss: 1.5892... Generator Loss: 0.6131
Epoch 1/5... Discriminator Loss: 1.6369... Generator Loss: 0.5745
Epoch 1/5... Discriminator Loss: 1.5439... Generator Loss: 0.6377
Epoch 1/5... Discriminator Loss: 1.5650... Generator Loss: 0.5977

Epoch 1/5... Discriminator Loss: 1.7046... Generator Loss: 0.5160
Epoch 1/5... Discriminator Loss: 1.5865... Generator Loss: 0.5771
Epoch 1/5... Discriminator Loss: 1.5817... Generator Loss: 0.5501
Epoch 1/5... Discriminator Loss: 1.5087... Generator Loss: 0.5750
Epoch 1/5... Discriminator Loss: 1.6477... Generator Loss: 0.5327
Epoch 1/5... Discriminator Loss: 1.5460... Generator Loss: 0.5528
Epoch 1/5... Discriminator Loss: 1.6487... Generator Loss: 0.6471
Epoch 1/5... Discriminator Loss: 1.6338... Generator Loss: 0.5165
Epoch 1/5... Discriminator Loss: 1.6565... Generator Loss: 0.5809
Epoch 1/5... Discriminator Loss: 1.5454... Generator Loss: 0.7024

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.